A novel reinforcement learning-based adaptive neural network (NN) controller, also referred as the adaptive-critic NN controller, is developed to deliver a desired tracking performance for a class of non-strict feedback nonlinear discrete-time systems in the presence of bounded and unknown disturbances. The adaptive critic NN controller architecture includes a critic NN and two action NNs. The critic NN approximates certain strategic utility function whereas the action neural networks are used to minimize both the strategic utility function and the unknown dynamics estimation errors. The NN weights are tuned online so as to minimize certain performance index. By using gradient descent-based novel weight updating rules, the uniformly ultimate boundedness (UUB) of the closed-loop tracking error and weight estimates is shown.
J. Sarangapani et al., "Neural Network-Based Control of Nonlinear Discrete-Time Systems in Non-Strict Form," Proceedings of the 44th IEEE Conference on Decision and Control and the European Control Conference (2005, Seville, Spain), Institute of Electrical and Electronics Engineers (IEEE), Jan 2005.
The definitive version is available at http://dx.doi.org/10.1109/CDC.2005.1582551
44th IEEE Conference on Decision and Control and the European Control Conference (2005: Dec. 12-15, Seville, Spain)
Electrical and Computer Engineering
Keywords and Phrases
Adaptive Neural Network (NN) Controller; Adaptive-Critic NN Controller; Non-Strict Feedback Nonlinear Discrete-Time Systems; Uniformly Ultimate Boundedness (UUB)
Article - Conference proceedings
© 2005 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.